kordoc 3.8.0 → 3.8.1

package/dist/index.js

@@ -19,7 +19,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-QV25HMU7.js";
+} from "./chunk-PELBIL4K.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
@@ -279,9 +279,7 @@ async function detectZipFormat(buffer) {
 }
 
 // src/hwpx/parser.ts
-import JSZip2 from "jszip";
-import { inflateRawSync } from "zlib";
-import { DOMParser } from "@xmldom/xmldom";
+import JSZip3 from "jszip";
 
 // src/hwpx/com-fallback.ts
 import { execFileSync } from "child_process";
@@ -390,6 +388,245 @@ function comResultToParseResult(pages, pageCount, warnings) {
   };
 }
 
+// src/hwpx/parser-shared.ts
+import { DOMParser } from "@xmldom/xmldom";
+var MAX_DECOMPRESS_SIZE = 100 * 1024 * 1024;
+var MAX_ZIP_ENTRIES = 500;
+function clampSpan(val, max) {
+  return Math.max(1, Math.min(val, max));
+}
+var MAX_XML_DEPTH = 200;
+function createSectionShared() {
+  return { numState: /* @__PURE__ */ new Map(), pageText: { headers: [], footers: [] }, track: { deleteDepth: 0, warned: false } };
+}
+function createXmlParser(warnings) {
+  return new DOMParser({
+    onError(level, msg2) {
+      if (level === "fatalError") throw new KordocError(`XML \uD30C\uC2F1 \uC2E4\uD328: ${msg2}`);
+      warnings?.push({ code: "MALFORMED_XML", message: `XML ${level === "warn" ? "\uACBD\uACE0" : "\uC624\uB958"}: ${msg2}` });
+    }
+  });
+}
+function applyPageText(blocks, shared) {
+  const { headers, footers } = shared.pageText;
+  if (headers.length > 0) {
+    blocks.unshift(...headers.map((t) => ({ type: "paragraph", text: t, pageNumber: 1 })));
+  }
+  if (footers.length > 0) {
+    blocks.push(...footers.map((t) => ({ type: "paragraph", text: t })));
+  }
+}
+function findChildByLocalName(parent, name) {
+  const children = parent.childNodes;
+  if (!children) return null;
+  for (let i = 0; i < children.length; i++) {
+    const ch = children[i];
+    if (ch.nodeType !== 1) continue;
+    const tag = (ch.tagName || ch.localName || "").replace(/^[^:]+:/, "");
+    if (tag === name) return ch;
+  }
+  return null;
+}
+function extractTextFromNode(node) {
+  let result = "";
+  const children = node.childNodes;
+  if (!children) return result;
+  for (let i = 0; i < children.length; i++) {
+    const child = children[i];
+    if (child.nodeType === 3) result += child.textContent || "";
+    else if (child.nodeType === 1) result += extractTextFromNode(child);
+  }
+  return result.trim();
+}
+
+// src/hwpx/styles.ts
+async function extractHwpxStyles(zip, decompressed) {
+  const result = {
+    charProperties: /* @__PURE__ */ new Map(),
+    styles: /* @__PURE__ */ new Map(),
+    numberings: /* @__PURE__ */ new Map(),
+    bullets: /* @__PURE__ */ new Map(),
+    paraHeadings: /* @__PURE__ */ new Map()
+  };
+  const headerPaths = ["Contents/header.xml", "header.xml", "Contents/head.xml", "head.xml"];
+  for (const hp of headerPaths) {
+    const hpLower = hp.toLowerCase();
+    const file = zip.file(hp) || Object.values(zip.files).find((f) => f.name.toLowerCase() === hpLower) || null;
+    if (!file) continue;
+    try {
+      const xml = await file.async("text");
+      if (decompressed) {
+        decompressed.total += xml.length * 2;
+        if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
+      }
+      const parser = createXmlParser();
+      const doc = parser.parseFromString(stripDtd(xml), "text/xml");
+      if (!doc.documentElement) continue;
+      parseCharProperties(doc, result.charProperties);
+      parseStyleElements(doc, result.styles);
+      const domDoc = doc;
+      parseNumberings(domDoc, result.numberings);
+      parseBullets(domDoc, result.bullets);
+      parseParaHeadings(domDoc, result.paraHeadings);
+      break;
+    } catch {
+      continue;
+    }
+  }
+  return result;
+}
+function parseCharProperties(doc, map) {
+  const tagNames = ["hh:charPr", "charPr", "hp:charPr"];
+  for (const tagName of tagNames) {
+    const elements = doc.getElementsByTagName(tagName);
+    for (let i = 0; i < elements.length; i++) {
+      const el = elements[i];
+      const id = el.getAttribute("id") || el.getAttribute("IDRef") || "";
+      if (!id) continue;
+      const prop = {};
+      const height = el.getAttribute("height");
+      if (height) {
+        const parsedHeight = parseInt(height, 10);
+        if (!isNaN(parsedHeight) && parsedHeight > 0) {
+          prop.fontSize = parsedHeight / 100;
+        }
+      }
+      const bold = el.getAttribute("bold");
+      if (bold === "true" || bold === "1") prop.bold = true;
+      const italic = el.getAttribute("italic");
+      if (italic === "true" || italic === "1") prop.italic = true;
+      const fontFaces = el.getElementsByTagName("*");
+      for (let j = 0; j < fontFaces.length; j++) {
+        const ff = fontFaces[j];
+        const localTag = (ff.tagName || "").replace(/^[^:]+:/, "");
+        if (localTag === "fontface" || localTag === "fontRef") {
+          const face = ff.getAttribute("face") || ff.getAttribute("FontFace");
+          if (face) {
+            prop.fontName = face;
+            break;
+          }
+        }
+      }
+      map.set(id, prop);
+    }
+  }
+}
+function parseStyleElements(doc, map) {
+  const tagNames = ["hh:style", "style", "hp:style"];
+  for (const tagName of tagNames) {
+    const elements = doc.getElementsByTagName(tagName);
+    for (let i = 0; i < elements.length; i++) {
+      const el = elements[i];
+      const id = el.getAttribute("id") || el.getAttribute("IDRef") || String(i);
+      const name = el.getAttribute("name") || el.getAttribute("engName") || "";
+      const charPrId = el.getAttribute("charPrIDRef") || void 0;
+      const paraPrId = el.getAttribute("paraPrIDRef") || void 0;
+      map.set(id, { name, charPrId, paraPrId });
+    }
+  }
+}
+function parseNumberings(doc, map) {
+  const tagNames = ["hh:numbering", "numbering"];
+  for (const tagName of tagNames) {
+    const elements = doc.getElementsByTagName(tagName);
+    for (let i = 0; i < elements.length; i++) {
+      const el = elements[i];
+      const id = el.getAttribute("id") || "";
+      if (!id) continue;
+      const def = { heads: /* @__PURE__ */ new Map() };
+      const children = el.childNodes;
+      for (let j = 0; j < children.length; j++) {
+        const ch = children[j];
+        if (ch.nodeType !== 1) continue;
+        const tag = (ch.tagName || ch.localName || "").replace(/^[^:]+:/, "");
+        if (tag !== "paraHead") continue;
+        const level = parseInt(ch.getAttribute("level") || "", 10);
+        if (isNaN(level) || level < 1 || level > 10) continue;
+        const start = parseInt(ch.getAttribute("start") || "1", 10);
+        def.heads.set(level, {
+          numFormat: ch.getAttribute("numFormat") || "DIGIT",
+          text: ch.textContent || "",
+          start: isNaN(start) ? 1 : start
+        });
+      }
+      if (def.heads.size > 0) map.set(id, def);
+    }
+    if (map.size > 0) break;
+  }
+}
+function parseBullets(doc, map) {
+  const tagNames = ["hh:bullet", "bullet"];
+  for (const tagName of tagNames) {
+    const elements = doc.getElementsByTagName(tagName);
+    for (let i = 0; i < elements.length; i++) {
+      const el = elements[i];
+      const id = el.getAttribute("id") || "";
+      const char = el.getAttribute("char") || "";
+      if (id && char) map.set(id, char);
+    }
+    if (map.size > 0) break;
+  }
+}
+function parseParaHeadings(doc, map) {
+  const tagNames = ["hh:paraPr", "paraPr"];
+  for (const tagName of tagNames) {
+    const elements = doc.getElementsByTagName(tagName);
+    for (let i = 0; i < elements.length; i++) {
+      const el = elements[i];
+      const id = el.getAttribute("id") || "";
+      if (!id) continue;
+      const heading = findChildByLocalName(el, "heading");
+      if (!heading) continue;
+      const type = heading.getAttribute("type") || "NONE";
+      if (type !== "NUMBER" && type !== "BULLET" && type !== "OUTLINE") continue;
+      const level = parseInt(heading.getAttribute("level") || "0", 10);
+      map.set(id, {
+        type,
+        idRef: heading.getAttribute("idRef") || "0",
+        level: isNaN(level) ? 0 : Math.max(0, Math.min(level, 9))
+      });
+    }
+    if (map.size > 0) break;
+  }
+}
+function detectHwpxHeadings(blocks, styleMap) {
+  if (blocks.some((b) => b.type === "heading")) return;
+  let baseFontSize = 0;
+  const sizeFreq = /* @__PURE__ */ new Map();
+  for (const b of blocks) {
+    if (b.style?.fontSize) {
+      sizeFreq.set(b.style.fontSize, (sizeFreq.get(b.style.fontSize) || 0) + 1);
+    }
+  }
+  let maxCount = 0;
+  for (const [size, count] of sizeFreq) {
+    if (count > maxCount) {
+      maxCount = count;
+      baseFontSize = size;
+    }
+  }
+  for (const block of blocks) {
+    if (block.type !== "paragraph" || !block.text) continue;
+    const text = block.text.trim();
+    if (text.length === 0 || text.length > 200 || /^\d+$/.test(text)) continue;
+    let level = 0;
+    if (baseFontSize > 0 && block.style?.fontSize) {
+      const ratio = block.style.fontSize / baseFontSize;
+      if (ratio >= HEADING_RATIO_H1) level = 1;
+      else if (ratio >= HEADING_RATIO_H2) level = 2;
+      else if (ratio >= HEADING_RATIO_H3) level = 3;
+    }
+    const compactText = text.replace(/\s+/g, "");
+    if (/^제\d+[조장절편]/.test(compactText) && text.length <= 50) {
+      if (level === 0) level = 3;
+    }
+    if (level > 0) {
+      block.type = "heading";
+      block.level = level;
+    }
+  }
+}
+
 // src/hwpx/equation.ts
 var CONVERT_MAP = {
   TIMES: "\\times",
@@ -829,629 +1066,101 @@ function hmlToLatex(hmlEqStr) {
   return out;
 }
 
-// src/hwpx/parser.ts
-var MAX_DECOMPRESS_SIZE = 100 * 1024 * 1024;
-var MAX_ZIP_ENTRIES = 500;
-function clampSpan(val, max) {
-  return Math.max(1, Math.min(val, max));
-}
-var MAX_XML_DEPTH = 200;
-function createSectionShared() {
-  return { numState: /* @__PURE__ */ new Map(), pageText: { headers: [], footers: [] }, track: { deleteDepth: 0, warned: false } };
-}
-function createXmlParser(warnings) {
-  return new DOMParser({
-    onError(level, msg2) {
-      if (level === "fatalError") throw new KordocError(`XML \uD30C\uC2F1 \uC2E4\uD328: ${msg2}`);
-      warnings?.push({ code: "MALFORMED_XML", message: `XML ${level === "warn" ? "\uACBD\uACE0" : "\uC624\uB958"}: ${msg2}` });
-    }
-  });
-}
-async function extractHwpxStyles(zip, decompressed) {
-  const result = {
-    charProperties: /* @__PURE__ */ new Map(),
-    styles: /* @__PURE__ */ new Map(),
-    numberings: /* @__PURE__ */ new Map(),
-    bullets: /* @__PURE__ */ new Map(),
-    paraHeadings: /* @__PURE__ */ new Map()
-  };
-  const headerPaths = ["Contents/header.xml", "header.xml", "Contents/head.xml", "head.xml"];
-  for (const hp of headerPaths) {
-    const hpLower = hp.toLowerCase();
-    const file = zip.file(hp) || Object.values(zip.files).find((f) => f.name.toLowerCase() === hpLower) || null;
-    if (!file) continue;
-    try {
-      const xml = await file.async("text");
-      if (decompressed) {
-        decompressed.total += xml.length * 2;
-        if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-      }
-      const parser = createXmlParser();
-      const doc = parser.parseFromString(stripDtd(xml), "text/xml");
-      if (!doc.documentElement) continue;
-      parseCharProperties(doc, result.charProperties);
-      parseStyleElements(doc, result.styles);
-      const domDoc = doc;
-      parseNumberings(domDoc, result.numberings);
-      parseBullets(domDoc, result.bullets);
-      parseParaHeadings(domDoc, result.paraHeadings);
-      break;
-    } catch {
-      continue;
-    }
-  }
-  return result;
-}
-function parseCharProperties(doc, map) {
-  const tagNames = ["hh:charPr", "charPr", "hp:charPr"];
-  for (const tagName of tagNames) {
-    const elements = doc.getElementsByTagName(tagName);
-    for (let i = 0; i < elements.length; i++) {
-      const el = elements[i];
-      const id = el.getAttribute("id") || el.getAttribute("IDRef") || "";
-      if (!id) continue;
-      const prop = {};
-      const height = el.getAttribute("height");
-      if (height) {
-        const parsedHeight = parseInt(height, 10);
-        if (!isNaN(parsedHeight) && parsedHeight > 0) {
-          prop.fontSize = parsedHeight / 100;
-        }
-      }
-      const bold = el.getAttribute("bold");
-      if (bold === "true" || bold === "1") prop.bold = true;
-      const italic = el.getAttribute("italic");
-      if (italic === "true" || italic === "1") prop.italic = true;
-      const fontFaces = el.getElementsByTagName("*");
-      for (let j = 0; j < fontFaces.length; j++) {
-        const ff = fontFaces[j];
-        const localTag = (ff.tagName || "").replace(/^[^:]+:/, "");
-        if (localTag === "fontface" || localTag === "fontRef") {
-          const face = ff.getAttribute("face") || ff.getAttribute("FontFace");
-          if (face) {
-            prop.fontName = face;
-            break;
-          }
-        }
-      }
-      map.set(id, prop);
-    }
-  }
-}
-function parseStyleElements(doc, map) {
-  const tagNames = ["hh:style", "style", "hp:style"];
-  for (const tagName of tagNames) {
-    const elements = doc.getElementsByTagName(tagName);
-    for (let i = 0; i < elements.length; i++) {
-      const el = elements[i];
-      const id = el.getAttribute("id") || el.getAttribute("IDRef") || String(i);
-      const name = el.getAttribute("name") || el.getAttribute("engName") || "";
-      const charPrId = el.getAttribute("charPrIDRef") || void 0;
-      const paraPrId = el.getAttribute("paraPrIDRef") || void 0;
-      map.set(id, { name, charPrId, paraPrId });
-    }
-  }
-}
-function parseNumberings(doc, map) {
-  const tagNames = ["hh:numbering", "numbering"];
-  for (const tagName of tagNames) {
-    const elements = doc.getElementsByTagName(tagName);
-    for (let i = 0; i < elements.length; i++) {
-      const el = elements[i];
-      const id = el.getAttribute("id") || "";
-      if (!id) continue;
-      const def = { heads: /* @__PURE__ */ new Map() };
-      const children = el.childNodes;
-      for (let j = 0; j < children.length; j++) {
-        const ch = children[j];
-        if (ch.nodeType !== 1) continue;
-        const tag = (ch.tagName || ch.localName || "").replace(/^[^:]+:/, "");
-        if (tag !== "paraHead") continue;
-        const level = parseInt(ch.getAttribute("level") || "", 10);
-        if (isNaN(level) || level < 1 || level > 10) continue;
-        const start = parseInt(ch.getAttribute("start") || "1", 10);
-        def.heads.set(level, {
-          numFormat: ch.getAttribute("numFormat") || "DIGIT",
-          text: ch.textContent || "",
-          start: isNaN(start) ? 1 : start
-        });
-      }
-      if (def.heads.size > 0) map.set(id, def);
-    }
-    if (map.size > 0) break;
-  }
-}
-function parseBullets(doc, map) {
-  const tagNames = ["hh:bullet", "bullet"];
-  for (const tagName of tagNames) {
-    const elements = doc.getElementsByTagName(tagName);
-    for (let i = 0; i < elements.length; i++) {
-      const el = elements[i];
-      const id = el.getAttribute("id") || "";
-      const char = el.getAttribute("char") || "";
-      if (id && char) map.set(id, char);
-    }
-    if (map.size > 0) break;
-  }
-}
-function parseParaHeadings(doc, map) {
-  const tagNames = ["hh:paraPr", "paraPr"];
-  for (const tagName of tagNames) {
-    const elements = doc.getElementsByTagName(tagName);
-    for (let i = 0; i < elements.length; i++) {
-      const el = elements[i];
-      const id = el.getAttribute("id") || "";
-      if (!id) continue;
-      const heading = findChildByLocalName(el, "heading");
-      if (!heading) continue;
-      const type = heading.getAttribute("type") || "NONE";
-      if (type !== "NUMBER" && type !== "BULLET" && type !== "OUTLINE") continue;
-      const level = parseInt(heading.getAttribute("level") || "0", 10);
-      map.set(id, {
-        type,
-        idRef: heading.getAttribute("idRef") || "0",
-        level: isNaN(level) ? 0 : Math.max(0, Math.min(level, 9))
-      });
-    }
-    if (map.size > 0) break;
-  }
-}
-var HANGUL_SYLLABLE_SEQ = "\uAC00\uB098\uB2E4\uB77C\uB9C8\uBC14\uC0AC\uC544\uC790\uCC28\uCE74\uD0C0\uD30C\uD558";
-var HANGUL_JAMO_SEQ = "\u3131\u3134\u3137\u3139\u3141\u3142\u3145\u3147\u3148\u314A\u314B\u314C\u314D\u314E";
-function toRoman(n) {
-  if (n <= 0 || n >= 4e3) return String(n);
-  const table = [
-    [1e3, "M"],
-    [900, "CM"],
-    [500, "D"],
-    [400, "CD"],
-    [100, "C"],
-    [90, "XC"],
-    [50, "L"],
-    [40, "XL"],
-    [10, "X"],
-    [9, "IX"],
-    [5, "V"],
-    [4, "IV"],
-    [1, "I"]
-  ];
-  let out = "";
-  for (const [v, s] of table) {
-    while (n >= v) {
-      out += s;
-      n -= v;
-    }
-  }
-  return out;
-}
-function formatHeadNumber(n, numFormat) {
-  if (n <= 0) n = 1;
-  switch (numFormat) {
-    case "DIGIT":
-      return String(n);
-    case "CIRCLED_DIGIT":
-      return n <= 20 ? String.fromCodePoint(9312 + n - 1) : `(${n})`;
-    case "HANGUL_SYLLABLE":
-      return HANGUL_SYLLABLE_SEQ[(n - 1) % HANGUL_SYLLABLE_SEQ.length];
-    case "CIRCLED_HANGUL_SYLLABLE":
-      return n <= 14 ? String.fromCodePoint(12910 + n - 1) : HANGUL_SYLLABLE_SEQ[(n - 1) % 14];
-    case "HANGUL_JAMO":
-      return HANGUL_JAMO_SEQ[(n - 1) % HANGUL_JAMO_SEQ.length];
-    case "CIRCLED_HANGUL_JAMO":
-      return n <= 14 ? String.fromCodePoint(12896 + n - 1) : HANGUL_JAMO_SEQ[(n - 1) % 14];
-    case "LATIN_CAPITAL":
-      return String.fromCharCode(65 + (n - 1) % 26);
-    case "LATIN_SMALL":
-      return String.fromCharCode(97 + (n - 1) % 26);
-    case "CIRCLED_LATIN_CAPITAL":
-      return n <= 26 ? String.fromCodePoint(9398 + n - 1) : String.fromCharCode(65 + (n - 1) % 26);
-    case "CIRCLED_LATIN_SMALL":
-      return n <= 26 ? String.fromCodePoint(9424 + n - 1) : String.fromCharCode(97 + (n - 1) % 26);
-    case "ROMAN_CAPITAL":
-      return toRoman(n);
-    case "ROMAN_SMALL":
-      return toRoman(n).toLowerCase();
-    default:
-      return String(n);
-  }
-}
-function resolveParaHeading(paraEl, ctx) {
-  const sm = ctx.styleMap;
-  if (!sm) return null;
-  const prId = paraEl.getAttribute("paraPrIDRef");
-  if (!prId) return null;
-  const ref = sm.paraHeadings.get(prId);
-  if (!ref) return null;
-  if (ref.type === "BULLET") {
-    const char = sm.bullets.get(ref.idRef);
-    return char ? { prefix: char } : null;
-  }
-  const numId = ref.type === "OUTLINE" ? ctx.outlineNumId || "1" : ref.idRef;
-  const level = Math.min(ref.level + 1, 10);
-  const headingLevel = ref.type === "OUTLINE" ? Math.min(ref.level + 1, 6) : void 0;
-  const numDef = sm.numberings.get(numId);
-  if (!numDef) return headingLevel ? { headingLevel } : null;
-  let counters = ctx.shared.numState.get(numId);
-  if (!counters) {
-    counters = new Array(11).fill(0);
-    ctx.shared.numState.set(numId, counters);
-  }
-  const head = numDef.heads.get(level);
-  counters[level] = counters[level] === 0 ? head?.start ?? 1 : counters[level] + 1;
-  for (let l = level + 1; l <= 10; l++) counters[l] = 0;
-  const fmtText = head?.text?.trim() || `^${level}.`;
-  const prefix = fmtText.replace(/\^(10|[1-9])/g, (_, d) => {
-    const lv = parseInt(d, 10);
-    const refHead = numDef.heads.get(lv);
-    const n = counters[lv] || refHead?.start || 1;
-    return formatHeadNumber(n, refHead?.numFormat || "DIGIT");
-  });
-  return { prefix, headingLevel };
-}
-async function parseHwpxDocument(buffer, options) {
-  precheckZipSize(buffer, MAX_DECOMPRESS_SIZE, MAX_ZIP_ENTRIES);
-  let zip;
-  try {
-    zip = await JSZip2.loadAsync(buffer);
-  } catch {
-    return extractFromBrokenZip(buffer);
-  }
-  const actualEntryCount = Object.keys(zip.files).length;
-  if (actualEntryCount > MAX_ZIP_ENTRIES) {
-    throw new KordocError("ZIP \uC5D4\uD2B8\uB9AC \uC218 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-  }
-  const manifestFile = zip.file("META-INF/manifest.xml");
-  if (manifestFile) {
-    const manifestXml = await manifestFile.async("text");
-    if (isEncryptedHwpx(manifestXml)) {
-      if (isComFallbackAvailable() && options?.filePath) {
-        const { pages, pageCount, warnings: warnings2 } = extractTextViaCom(options.filePath);
-        if (pages.some((p) => p && p.trim().length > 0)) {
-          return comResultToParseResult(pages, pageCount, warnings2);
-        }
-      }
-      throw new KordocError("DRM \uC554\uD638\uD654\uB41C HWPX \uD30C\uC77C\uC785\uB2C8\uB2E4. Windows + \uD55C\uCEF4 \uC624\uD53C\uC2A4 \uC124\uCE58 \uC2DC \uC790\uB3D9 \uCD94\uCD9C\uB429\uB2C8\uB2E4.");
-    }
-  }
-  const decompressed = { total: 0 };
-  const metadata = {};
-  await extractHwpxMetadata(zip, metadata, decompressed);
-  const styleMap = await extractHwpxStyles(zip, decompressed);
-  const warnings = [];
-  const sectionPaths = await resolveSectionPaths(zip);
-  if (sectionPaths.length === 0) throw new KordocError("HWPX\uC5D0\uC11C \uC139\uC158 \uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
-  metadata.pageCount = sectionPaths.length;
-  const pageFilter = options?.pages ? parsePageRange(options.pages, sectionPaths.length) : null;
-  const totalTarget = pageFilter ? pageFilter.size : sectionPaths.length;
-  const blocks = [];
-  const shared = createSectionShared();
-  let parsedSections = 0;
-  for (let si = 0; si < sectionPaths.length; si++) {
-    if (pageFilter && !pageFilter.has(si + 1)) continue;
-    const file = zip.file(sectionPaths[si]);
-    if (!file) continue;
-    try {
-      const xml = await file.async("text");
-      decompressed.total += xml.length * 2;
-      if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-      blocks.push(...parseSectionXml(xml, styleMap, warnings, si + 1, shared));
-      parsedSections++;
-      options?.onProgress?.(parsedSections, totalTarget);
-    } catch (secErr) {
-      if (secErr instanceof KordocError) throw secErr;
-      warnings.push({ page: si + 1, message: `\uC139\uC158 ${si + 1} \uD30C\uC2F1 \uC2E4\uD328: ${secErr instanceof Error ? secErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
-    }
-  }
-  applyPageText(blocks, shared);
-  const images = await extractImagesFromZip(zip, blocks, decompressed, warnings);
-  detectHwpxHeadings(blocks, styleMap);
-  const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
-  const markdown = blocksToMarkdown(blocks);
-  return { markdown, blocks, metadata, outline: outline.length > 0 ? outline : void 0, warnings: warnings.length > 0 ? warnings : void 0, images: images.length > 0 ? images : void 0 };
-}
-function applyPageText(blocks, shared) {
-  const { headers, footers } = shared.pageText;
-  if (headers.length > 0) {
-    blocks.unshift(...headers.map((t) => ({ type: "paragraph", text: t, pageNumber: 1 })));
-  }
-  if (footers.length > 0) {
-    blocks.push(...footers.map((t) => ({ type: "paragraph", text: t })));
-  }
-}
-function imageExtToMime(ext) {
-  switch (ext.toLowerCase()) {
-    case "jpg":
-    case "jpeg":
-      return "image/jpeg";
-    case "png":
-      return "image/png";
-    case "gif":
-      return "image/gif";
-    case "bmp":
-      return "image/bmp";
-    case "tif":
-    case "tiff":
-      return "image/tiff";
-    case "wmf":
-      return "image/wmf";
-    case "emf":
-      return "image/emf";
-    case "svg":
-      return "image/svg+xml";
-    default:
-      return "application/octet-stream";
-  }
-}
-function mimeToExt(mime) {
-  if (mime.includes("jpeg")) return "jpg";
-  if (mime.includes("png")) return "png";
-  if (mime.includes("gif")) return "gif";
-  if (mime.includes("bmp")) return "bmp";
-  if (mime.includes("tiff")) return "tif";
-  if (mime.includes("wmf")) return "wmf";
-  if (mime.includes("emf")) return "emf";
-  if (mime.includes("svg")) return "svg";
-  return "bin";
-}
-function collectImageBlocks(blocks, out, ownerCell, depth = 0) {
-  if (depth > MAX_XML_DEPTH) return;
-  for (const block of blocks) {
-    if (block.type === "image") {
-      out.push({ block, ownerCell });
-    } else if (block.type === "table" && block.table) {
-      for (const row of block.table.cells) {
-        for (const cell of row) {
-          if (cell.blocks?.length) collectImageBlocks(cell.blocks, out, cell, depth + 1);
-        }
-      }
-    }
-  }
-}
-async function extractImagesFromZip(zip, blocks, decompressed, warnings) {
-  const images = [];
-  let imageIndex = 0;
-  const imageBlocks = [];
-  collectImageBlocks(blocks, imageBlocks);
-  const resolved = /* @__PURE__ */ new Map();
-  for (const { block, ownerCell } of imageBlocks) {
-    if (block.type !== "image" || !block.text) continue;
-    const ref = block.text;
-    let img = resolved.get(ref);
-    if (img === void 0) {
-      img = null;
-      const candidates = [
-        `BinData/${ref}`,
-        `Contents/BinData/${ref}`,
-        ref
-        // 절대 경로일 수도 있음
-      ];
-      let resolvedPath = null;
-      if (!ref.includes(".")) {
-        const prefixes = [`BinData/${ref}`, `Contents/BinData/${ref}`];
-        for (const prefix of prefixes) {
-          const match = zip.file(new RegExp(`^${prefix.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}\\.[a-zA-Z0-9]+$`));
-          if (match.length > 0) {
-            resolvedPath = match[0].name;
-            break;
-          }
-        }
-      }
-      const allCandidates = resolvedPath ? [resolvedPath, ...candidates] : candidates;
-      for (const path of allCandidates) {
-        if (isPathTraversal(path)) continue;
-        const file = zip.file(path);
-        if (!file) continue;
-        try {
-          const data = await file.async("uint8array");
-          decompressed.total += data.length;
-          if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-          const ext = path.includes(".") ? path.split(".").pop() || "png" : "png";
-          const mimeType = imageExtToMime(ext);
-          imageIndex++;
-          const filename = `image_${String(imageIndex).padStart(3, "0")}.${mimeToExt(mimeType)}`;
-          img = { filename, data, mimeType };
-          images.push(img);
-          break;
-        } catch (err) {
-          if (err instanceof KordocError) throw err;
-        }
-      }
-      if (!img) warnings?.push({ page: block.pageNumber, message: `\uC774\uBBF8\uC9C0 \uD30C\uC77C \uC5C6\uC74C: ${ref}`, code: "SKIPPED_IMAGE" });
-      resolved.set(ref, img);
-    }
-    if (!img) {
-      block.type = "paragraph";
-      block.text = `[\uC774\uBBF8\uC9C0: ${ref}]`;
-      if (ownerCell) ownerCell.text = ownerCell.text.replace(`![image](${ref})`, `[\uC774\uBBF8\uC9C0: ${ref}]`);
-      continue;
-    }
-    block.text = img.filename;
-    block.imageData = { data: img.data, mimeType: img.mimeType, filename: ref };
-    if (ownerCell) ownerCell.text = ownerCell.text.replace(`![image](${ref})`, `![image](${img.filename})`);
-  }
-  return images;
-}
-async function extractHwpxMetadata(zip, metadata, decompressed) {
-  try {
-    const metaPaths = ["meta.xml", "META-INF/meta.xml", "docProps/core.xml"];
-    for (const mp of metaPaths) {
-      const file = zip.file(mp) || Object.values(zip.files).find((f) => f.name.toLowerCase() === mp.toLowerCase()) || null;
-      if (!file) continue;
-      const xml = await file.async("text");
-      if (decompressed) {
-        decompressed.total += xml.length * 2;
-        if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-      }
-      parseDublinCoreMetadata(xml, metadata);
-      if (metadata.title || metadata.author) return;
-    }
-  } catch {
-  }
-}
-function parseDublinCoreMetadata(xml, metadata) {
-  const parser = createXmlParser();
-  const doc = parser.parseFromString(stripDtd(xml), "text/xml");
-  if (!doc.documentElement) return;
-  const getText = (tagNames) => {
-    for (const tag of tagNames) {
-      const els = doc.getElementsByTagName(tag);
-      if (els.length > 0) {
-        const text = els[0].textContent?.trim();
-        if (text) return text;
-      }
-    }
-    return void 0;
-  };
-  metadata.title = metadata.title || getText(["dc:title", "title"]);
-  metadata.author = metadata.author || getText(["dc:creator", "creator", "cp:lastModifiedBy"]);
-  metadata.description = metadata.description || getText(["dc:description", "description", "dc:subject", "subject"]);
-  metadata.createdAt = metadata.createdAt || getText(["dcterms:created", "meta:creation-date"]);
-  metadata.modifiedAt = metadata.modifiedAt || getText(["dcterms:modified", "meta:date"]);
-  const keywords = getText(["dc:keyword", "cp:keywords", "meta:keyword"]);
-  if (keywords && !metadata.keywords) {
-    metadata.keywords = keywords.split(/[,;]/).map((k) => k.trim()).filter(Boolean);
-  }
-}
-function extractFromBrokenZip(buffer) {
-  const data = new Uint8Array(buffer);
-  const view = new DataView(buffer);
-  let pos = 0;
-  const blocks = [];
-  const warnings = [
-    { code: "BROKEN_ZIP_RECOVERY", message: "\uC190\uC0C1\uB41C ZIP \uAD6C\uC870 \u2014 Local File Header \uAE30\uBC18 \uBCF5\uAD6C \uBAA8\uB4DC" }
-  ];
-  let totalDecompressed = 0;
-  let entryCount = 0;
-  let sectionNum = 0;
-  const shared = createSectionShared();
-  while (pos < data.length - 30) {
-    if (data[pos] !== 80 || data[pos + 1] !== 75 || data[pos + 2] !== 3 || data[pos + 3] !== 4) {
-      pos++;
-      while (pos < data.length - 30) {
-        if (data[pos] === 80 && data[pos + 1] === 75 && data[pos + 2] === 3 && data[pos + 3] === 4) break;
-        pos++;
-      }
-      continue;
-    }
-    if (++entryCount > MAX_ZIP_ENTRIES) break;
-    const method = view.getUint16(pos + 8, true);
-    const compSize = view.getUint32(pos + 18, true);
-    const nameLen = view.getUint16(pos + 26, true);
-    const extraLen = view.getUint16(pos + 28, true);
-    if (nameLen > 1024 || extraLen > 65535) {
-      pos += 30 + nameLen + extraLen;
-      continue;
-    }
-    const fileStart = pos + 30 + nameLen + extraLen;
-    if (fileStart + compSize > data.length) break;
-    if (compSize === 0 && method !== 0) {
-      pos = fileStart;
-      continue;
-    }
-    const nameBytes = data.slice(pos + 30, pos + 30 + nameLen);
-    const name = new TextDecoder().decode(nameBytes);
-    if (isPathTraversal(name)) {
-      pos = fileStart + compSize;
-      continue;
-    }
-    const fileData = data.slice(fileStart, fileStart + compSize);
-    pos = fileStart + compSize;
-    if (!name.toLowerCase().includes("section") || !name.endsWith(".xml")) continue;
-    try {
-      let content;
-      if (method === 0) {
-        content = new TextDecoder().decode(fileData);
-      } else if (method === 8) {
-        const decompressed = inflateRawSync(Buffer.from(fileData), { maxOutputLength: MAX_DECOMPRESS_SIZE });
-        content = new TextDecoder().decode(decompressed);
-      } else {
-        continue;
-      }
-      totalDecompressed += content.length * 2;
-      if (totalDecompressed > MAX_DECOMPRESS_SIZE) throw new KordocError("\uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC");
-      sectionNum++;
-      blocks.push(...parseSectionXml(content, void 0, warnings, sectionNum, shared));
-    } catch {
-      continue;
-    }
-  }
-  if (blocks.length === 0) throw new KordocError("\uC190\uC0C1\uB41C HWPX\uC5D0\uC11C \uC139\uC158 \uB370\uC774\uD130\uB97C \uBCF5\uAD6C\uD560 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
-  applyPageText(blocks, shared);
-  const markdown = blocksToMarkdown(blocks);
-  return { markdown, blocks, warnings: warnings.length > 0 ? warnings : void 0 };
-}
-async function resolveSectionPaths(zip) {
-  const manifestPaths = ["Contents/content.hpf", "content.hpf"];
-  for (const mp of manifestPaths) {
-    const mpLower = mp.toLowerCase();
-    const file = zip.file(mp) || Object.values(zip.files).find((f) => f.name.toLowerCase() === mpLower) || null;
-    if (!file) continue;
-    const xml = await file.async("text");
-    const paths = parseSectionPathsFromManifest(xml);
-    if (paths.length > 0) return paths;
-  }
-  const sectionFiles = zip.file(/[Ss]ection\d+\.xml$/);
-  return sectionFiles.map((f) => f.name).sort(compareSectionPaths);
-}
-function parseSectionPathsFromManifest(xml) {
-  const parser = createXmlParser();
-  const doc = parser.parseFromString(stripDtd(xml), "text/xml");
-  const items = doc.getElementsByTagName("opf:item");
-  const spine = doc.getElementsByTagName("opf:itemref");
-  const idToHref = /* @__PURE__ */ new Map();
-  for (let i = 0; i < items.length; i++) {
-    const item = items[i];
-    const id = item.getAttribute("id") || "";
-    const href = normalizeSectionHref(item.getAttribute("href") || "");
-    if (id && href) idToHref.set(id, href);
-  }
-  if (spine.length > 0) {
-    const ordered = [];
-    for (let i = 0; i < spine.length; i++) {
-      const href = idToHref.get(spine[i].getAttribute("idref") || "");
-      if (href) ordered.push(href);
+// src/hwpx/para-heading.ts
+var HANGUL_SYLLABLE_SEQ = "\uAC00\uB098\uB2E4\uB77C\uB9C8\uBC14\uC0AC\uC544\uC790\uCC28\uCE74\uD0C0\uD30C\uD558";
+var HANGUL_JAMO_SEQ = "\u3131\u3134\u3137\u3139\u3141\u3142\u3145\u3147\u3148\u314A\u314B\u314C\u314D\u314E";
+function toRoman(n) {
+  if (n <= 0 || n >= 4e3) return String(n);
+  const table = [
+    [1e3, "M"],
+    [900, "CM"],
+    [500, "D"],
+    [400, "CD"],
+    [100, "C"],
+    [90, "XC"],
+    [50, "L"],
+    [40, "XL"],
+    [10, "X"],
+    [9, "IX"],
+    [5, "V"],
+    [4, "IV"],
+    [1, "I"]
+  ];
+  let out = "";
+  for (const [v, s] of table) {
+    while (n >= v) {
+      out += s;
+      n -= v;
     }
-    if (ordered.length > 0) return ordered;
   }
-  return Array.from(idToHref.values()).sort(compareSectionPaths);
+  return out;
 }
-function detectHwpxHeadings(blocks, styleMap) {
-  if (blocks.some((b) => b.type === "heading")) return;
-  let baseFontSize = 0;
-  const sizeFreq = /* @__PURE__ */ new Map();
-  for (const b of blocks) {
-    if (b.style?.fontSize) {
-      sizeFreq.set(b.style.fontSize, (sizeFreq.get(b.style.fontSize) || 0) + 1);
-    }
+function formatHeadNumber(n, numFormat) {
+  if (n <= 0) n = 1;
+  switch (numFormat) {
+    case "DIGIT":
+      return String(n);
+    case "CIRCLED_DIGIT":
+      return n <= 20 ? String.fromCodePoint(9312 + n - 1) : `(${n})`;
+    case "HANGUL_SYLLABLE":
+      return HANGUL_SYLLABLE_SEQ[(n - 1) % HANGUL_SYLLABLE_SEQ.length];
+    case "CIRCLED_HANGUL_SYLLABLE":
+      return n <= 14 ? String.fromCodePoint(12910 + n - 1) : HANGUL_SYLLABLE_SEQ[(n - 1) % 14];
+    case "HANGUL_JAMO":
+      return HANGUL_JAMO_SEQ[(n - 1) % HANGUL_JAMO_SEQ.length];
+    case "CIRCLED_HANGUL_JAMO":
+      return n <= 14 ? String.fromCodePoint(12896 + n - 1) : HANGUL_JAMO_SEQ[(n - 1) % 14];
+    case "LATIN_CAPITAL":
+      return String.fromCharCode(65 + (n - 1) % 26);
+    case "LATIN_SMALL":
+      return String.fromCharCode(97 + (n - 1) % 26);
+    case "CIRCLED_LATIN_CAPITAL":
+      return n <= 26 ? String.fromCodePoint(9398 + n - 1) : String.fromCharCode(65 + (n - 1) % 26);
+    case "CIRCLED_LATIN_SMALL":
+      return n <= 26 ? String.fromCodePoint(9424 + n - 1) : String.fromCharCode(97 + (n - 1) % 26);
+    case "ROMAN_CAPITAL":
+      return toRoman(n);
+    case "ROMAN_SMALL":
+      return toRoman(n).toLowerCase();
+    default:
+      return String(n);
   }
-  let maxCount = 0;
-  for (const [size, count] of sizeFreq) {
-    if (count > maxCount) {
-      maxCount = count;
-      baseFontSize = size;
-    }
+}
+function resolveParaHeading(paraEl, ctx) {
+  const sm = ctx.styleMap;
+  if (!sm) return null;
+  const prId = paraEl.getAttribute("paraPrIDRef");
+  if (!prId) return null;
+  const ref = sm.paraHeadings.get(prId);
+  if (!ref) return null;
+  if (ref.type === "BULLET") {
+    const char = sm.bullets.get(ref.idRef);
+    return char ? { prefix: char } : null;
   }
-  for (const block of blocks) {
-    if (block.type !== "paragraph" || !block.text) continue;
-    const text = block.text.trim();
-    if (text.length === 0 || text.length > 200 || /^\d+$/.test(text)) continue;
-    let level = 0;
-    if (baseFontSize > 0 && block.style?.fontSize) {
-      const ratio = block.style.fontSize / baseFontSize;
-      if (ratio >= HEADING_RATIO_H1) level = 1;
-      else if (ratio >= HEADING_RATIO_H2) level = 2;
-      else if (ratio >= HEADING_RATIO_H3) level = 3;
-    }
-    const compactText = text.replace(/\s+/g, "");
-    if (/^제\d+[조장절편]/.test(compactText) && text.length <= 50) {
-      if (level === 0) level = 3;
-    }
-    if (level > 0) {
-      block.type = "heading";
-      block.level = level;
-    }
+  const numId = ref.type === "OUTLINE" ? ctx.outlineNumId || "1" : ref.idRef;
+  const level = Math.min(ref.level + 1, 10);
+  const headingLevel = ref.type === "OUTLINE" ? Math.min(ref.level + 1, 6) : void 0;
+  const numDef = sm.numberings.get(numId);
+  if (!numDef) return headingLevel ? { headingLevel } : null;
+  let counters = ctx.shared.numState.get(numId);
+  if (!counters) {
+    counters = new Array(11).fill(0);
+    ctx.shared.numState.set(numId, counters);
   }
+  const head = numDef.heads.get(level);
+  counters[level] = counters[level] === 0 ? head?.start ?? 1 : counters[level] + 1;
+  for (let l = level + 1; l <= 10; l++) counters[l] = 0;
+  const fmtText = head?.text?.trim() || `^${level}.`;
+  const prefix = fmtText.replace(/\^(10|[1-9])/g, (_, d) => {
+    const lv = parseInt(d, 10);
+    const refHead = numDef.heads.get(lv);
+    const n = counters[lv] || refHead?.start || 1;
+    return formatHeadNumber(n, refHead?.numFormat || "DIGIT");
+  });
+  return { prefix, headingLevel };
 }
+
+// src/hwpx/table-build.ts
 function buildTableWithCellMeta(state) {
   const table = buildTable(state.rows);
   if (state.caption) table.caption = state.caption;
@@ -1531,6 +1240,8 @@ function completeTable(newTable, tableStack, blocks, ctx) {
   }
   return parentTable;
 }
+
+// src/hwpx/section-walker.ts
 function parseSectionXml(xml, styleMap, warnings, sectionNum, shared) {
   const parser = createXmlParser(warnings);
   const doc = parser.parseFromString(stripDtd(xml), "text/xml");
@@ -2064,49 +1775,357 @@ function extractParagraphInfo(para, styleMap, ctx) {
           break;
       }
     }
-  };
-  walk(para);
-  const leaderIdx = text.indexOf("");
-  if (leaderIdx >= 0) text = text.substring(0, leaderIdx);
-  let cleanText = text.replace(/[ \t]+/g, " ").trim();
-  if (/^그림입니다\.?\s*원본\s*그림의\s*(이름|크기)/.test(cleanText)) cleanText = "";
-  cleanText = cleanText.replace(/그림입니다\.?\s*원본\s*그림의\s*(이름|크기)[^\n]*(\n[^\n]*원본\s*그림의\s*(이름|크기)[^\n]*)*/g, "").trim();
-  cleanText = cleanText.replace(/(?:모서리가 둥근 |둥근 )?(?:사각형|직사각형|정사각형|원|타원|삼각형|선|직선|곡선|화살표|오각형|육각형|팔각형|별|십자|구름|마름모|도넛|평행사변형|사다리꼴|개체|그리기\s?개체|묶음\s?개체|글상자|표|그림|OLE\s?개체)\s?입니다\.?/g, "").trim();
-  let style;
-  if (styleMap && charPrId) {
-    const charProp = styleMap.charProperties.get(charPrId);
-    if (charProp) {
-      style = {};
-      if (charProp.fontSize) style.fontSize = charProp.fontSize;
-      if (charProp.bold) style.bold = true;
-      if (charProp.italic) style.italic = true;
-      if (charProp.fontName) style.fontName = charProp.fontName;
-      if (!style.fontSize && !style.bold && !style.italic) style = void 0;
+  };
+  walk(para);
+  const leaderIdx = text.indexOf("");
+  if (leaderIdx >= 0) text = text.substring(0, leaderIdx);
+  let cleanText = text.replace(/[ \t]+/g, " ").trim();
+  if (/^그림입니다\.?\s*원본\s*그림의\s*(이름|크기)/.test(cleanText)) cleanText = "";
+  cleanText = cleanText.replace(/그림입니다\.?\s*원본\s*그림의\s*(이름|크기)[^\n]*(\n[^\n]*원본\s*그림의\s*(이름|크기)[^\n]*)*/g, "").trim();
+  cleanText = cleanText.replace(/(?:모서리가 둥근 |둥근 )?(?:사각형|직사각형|정사각형|원|타원|삼각형|선|직선|곡선|화살표|오각형|육각형|팔각형|별|십자|구름|마름모|도넛|평행사변형|사다리꼴|개체|그리기\s?개체|묶음\s?개체|글상자|표|그림|OLE\s?개체)\s?입니다\.?/g, "").trim();
+  let style;
+  if (styleMap && charPrId) {
+    const charProp = styleMap.charProperties.get(charPrId);
+    if (charProp) {
+      style = {};
+      if (charProp.fontSize) style.fontSize = charProp.fontSize;
+      if (charProp.bold) style.bold = true;
+      if (charProp.italic) style.italic = true;
+      if (charProp.fontName) style.fontName = charProp.fontName;
+      if (!style.fontSize && !style.bold && !style.italic) style = void 0;
+    }
+  }
+  return { text: cleanText, href, footnote, style };
+}
+
+// src/hwpx/images.ts
+function imageExtToMime(ext) {
+  switch (ext.toLowerCase()) {
+    case "jpg":
+    case "jpeg":
+      return "image/jpeg";
+    case "png":
+      return "image/png";
+    case "gif":
+      return "image/gif";
+    case "bmp":
+      return "image/bmp";
+    case "tif":
+    case "tiff":
+      return "image/tiff";
+    case "wmf":
+      return "image/wmf";
+    case "emf":
+      return "image/emf";
+    case "svg":
+      return "image/svg+xml";
+    default:
+      return "application/octet-stream";
+  }
+}
+function mimeToExt(mime) {
+  if (mime.includes("jpeg")) return "jpg";
+  if (mime.includes("png")) return "png";
+  if (mime.includes("gif")) return "gif";
+  if (mime.includes("bmp")) return "bmp";
+  if (mime.includes("tiff")) return "tif";
+  if (mime.includes("wmf")) return "wmf";
+  if (mime.includes("emf")) return "emf";
+  if (mime.includes("svg")) return "svg";
+  return "bin";
+}
+function collectImageBlocks(blocks, out, ownerCell, depth = 0) {
+  if (depth > MAX_XML_DEPTH) return;
+  for (const block of blocks) {
+    if (block.type === "image") {
+      out.push({ block, ownerCell });
+    } else if (block.type === "table" && block.table) {
+      for (const row of block.table.cells) {
+        for (const cell of row) {
+          if (cell.blocks?.length) collectImageBlocks(cell.blocks, out, cell, depth + 1);
+        }
+      }
+    }
+  }
+}
+async function extractImagesFromZip(zip, blocks, decompressed, warnings) {
+  const images = [];
+  let imageIndex = 0;
+  const imageBlocks = [];
+  collectImageBlocks(blocks, imageBlocks);
+  const resolved = /* @__PURE__ */ new Map();
+  for (const { block, ownerCell } of imageBlocks) {
+    if (block.type !== "image" || !block.text) continue;
+    const ref = block.text;
+    let img = resolved.get(ref);
+    if (img === void 0) {
+      img = null;
+      const candidates = [
+        `BinData/${ref}`,
+        `Contents/BinData/${ref}`,
+        ref
+        // 절대 경로일 수도 있음
+      ];
+      let resolvedPath = null;
+      if (!ref.includes(".")) {
+        const prefixes = [`BinData/${ref}`, `Contents/BinData/${ref}`];
+        for (const prefix of prefixes) {
+          const match = zip.file(new RegExp(`^${prefix.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}\\.[a-zA-Z0-9]+$`));
+          if (match.length > 0) {
+            resolvedPath = match[0].name;
+            break;
+          }
+        }
+      }
+      const allCandidates = resolvedPath ? [resolvedPath, ...candidates] : candidates;
+      for (const path of allCandidates) {
+        if (isPathTraversal(path)) continue;
+        const file = zip.file(path);
+        if (!file) continue;
+        try {
+          const data = await file.async("uint8array");
+          decompressed.total += data.length;
+          if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
+          const ext = path.includes(".") ? path.split(".").pop() || "png" : "png";
+          const mimeType = imageExtToMime(ext);
+          imageIndex++;
+          const filename = `image_${String(imageIndex).padStart(3, "0")}.${mimeToExt(mimeType)}`;
+          img = { filename, data, mimeType };
+          images.push(img);
+          break;
+        } catch (err) {
+          if (err instanceof KordocError) throw err;
+        }
+      }
+      if (!img) warnings?.push({ page: block.pageNumber, message: `\uC774\uBBF8\uC9C0 \uD30C\uC77C \uC5C6\uC74C: ${ref}`, code: "SKIPPED_IMAGE" });
+      resolved.set(ref, img);
+    }
+    if (!img) {
+      block.type = "paragraph";
+      block.text = `[\uC774\uBBF8\uC9C0: ${ref}]`;
+      if (ownerCell) ownerCell.text = ownerCell.text.replace(`![image](${ref})`, `[\uC774\uBBF8\uC9C0: ${ref}]`);
+      continue;
+    }
+    block.text = img.filename;
+    block.imageData = { data: img.data, mimeType: img.mimeType, filename: ref };
+    if (ownerCell) ownerCell.text = ownerCell.text.replace(`![image](${ref})`, `![image](${img.filename})`);
+  }
+  return images;
+}
+
+// src/hwpx/metadata.ts
+import JSZip2 from "jszip";
+
+// src/hwpx/zip-sections.ts
+import { inflateRawSync } from "zlib";
+function extractFromBrokenZip(buffer) {
+  const data = new Uint8Array(buffer);
+  const view = new DataView(buffer);
+  let pos = 0;
+  const blocks = [];
+  const warnings = [
+    { code: "BROKEN_ZIP_RECOVERY", message: "\uC190\uC0C1\uB41C ZIP \uAD6C\uC870 \u2014 Local File Header \uAE30\uBC18 \uBCF5\uAD6C \uBAA8\uB4DC" }
+  ];
+  let totalDecompressed = 0;
+  let entryCount = 0;
+  let sectionNum = 0;
+  const shared = createSectionShared();
+  while (pos < data.length - 30) {
+    if (data[pos] !== 80 || data[pos + 1] !== 75 || data[pos + 2] !== 3 || data[pos + 3] !== 4) {
+      pos++;
+      while (pos < data.length - 30) {
+        if (data[pos] === 80 && data[pos + 1] === 75 && data[pos + 2] === 3 && data[pos + 3] === 4) break;
+        pos++;
+      }
+      continue;
+    }
+    if (++entryCount > MAX_ZIP_ENTRIES) break;
+    const method = view.getUint16(pos + 8, true);
+    const compSize = view.getUint32(pos + 18, true);
+    const nameLen = view.getUint16(pos + 26, true);
+    const extraLen = view.getUint16(pos + 28, true);
+    if (nameLen > 1024 || extraLen > 65535) {
+      pos += 30 + nameLen + extraLen;
+      continue;
+    }
+    const fileStart = pos + 30 + nameLen + extraLen;
+    if (fileStart + compSize > data.length) break;
+    if (compSize === 0 && method !== 0) {
+      pos = fileStart;
+      continue;
+    }
+    const nameBytes = data.slice(pos + 30, pos + 30 + nameLen);
+    const name = new TextDecoder().decode(nameBytes);
+    if (isPathTraversal(name)) {
+      pos = fileStart + compSize;
+      continue;
+    }
+    const fileData = data.slice(fileStart, fileStart + compSize);
+    pos = fileStart + compSize;
+    if (!name.toLowerCase().includes("section") || !name.endsWith(".xml")) continue;
+    try {
+      let content;
+      if (method === 0) {
+        content = new TextDecoder().decode(fileData);
+      } else if (method === 8) {
+        const decompressed = inflateRawSync(Buffer.from(fileData), { maxOutputLength: MAX_DECOMPRESS_SIZE });
+        content = new TextDecoder().decode(decompressed);
+      } else {
+        continue;
+      }
+      totalDecompressed += content.length * 2;
+      if (totalDecompressed > MAX_DECOMPRESS_SIZE) throw new KordocError("\uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC");
+      sectionNum++;
+      blocks.push(...parseSectionXml(content, void 0, warnings, sectionNum, shared));
+    } catch {
+      continue;
     }
   }
-  return { text: cleanText, href, footnote, style };
+  if (blocks.length === 0) throw new KordocError("\uC190\uC0C1\uB41C HWPX\uC5D0\uC11C \uC139\uC158 \uB370\uC774\uD130\uB97C \uBCF5\uAD6C\uD560 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
+  applyPageText(blocks, shared);
+  const markdown = blocksToMarkdown(blocks);
+  return { markdown, blocks, warnings: warnings.length > 0 ? warnings : void 0 };
 }
-function findChildByLocalName(parent, name) {
-  const children = parent.childNodes;
-  if (!children) return null;
-  for (let i = 0; i < children.length; i++) {
-    const ch = children[i];
-    if (ch.nodeType !== 1) continue;
-    const tag = (ch.tagName || ch.localName || "").replace(/^[^:]+:/, "");
-    if (tag === name) return ch;
+async function resolveSectionPaths(zip) {
+  const manifestPaths = ["Contents/content.hpf", "content.hpf"];
+  for (const mp of manifestPaths) {
+    const mpLower = mp.toLowerCase();
+    const file = zip.file(mp) || Object.values(zip.files).find((f) => f.name.toLowerCase() === mpLower) || null;
+    if (!file) continue;
+    const xml = await file.async("text");
+    const paths = parseSectionPathsFromManifest(xml);
+    if (paths.length > 0) return paths;
   }
-  return null;
+  const sectionFiles = zip.file(/[Ss]ection\d+\.xml$/);
+  return sectionFiles.map((f) => f.name).sort(compareSectionPaths);
 }
-function extractTextFromNode(node) {
-  let result = "";
-  const children = node.childNodes;
-  if (!children) return result;
-  for (let i = 0; i < children.length; i++) {
-    const child = children[i];
-    if (child.nodeType === 3) result += child.textContent || "";
-    else if (child.nodeType === 1) result += extractTextFromNode(child);
+function parseSectionPathsFromManifest(xml) {
+  const parser = createXmlParser();
+  const doc = parser.parseFromString(stripDtd(xml), "text/xml");
+  const items = doc.getElementsByTagName("opf:item");
+  const spine = doc.getElementsByTagName("opf:itemref");
+  const idToHref = /* @__PURE__ */ new Map();
+  for (let i = 0; i < items.length; i++) {
+    const item = items[i];
+    const id = item.getAttribute("id") || "";
+    const href = normalizeSectionHref(item.getAttribute("href") || "");
+    if (id && href) idToHref.set(id, href);
   }
-  return result.trim();
+  if (spine.length > 0) {
+    const ordered = [];
+    for (let i = 0; i < spine.length; i++) {
+      const href = idToHref.get(spine[i].getAttribute("idref") || "");
+      if (href) ordered.push(href);
+    }
+    if (ordered.length > 0) return ordered;
+  }
+  return Array.from(idToHref.values()).sort(compareSectionPaths);
+}
+
+// src/hwpx/metadata.ts
+async function extractHwpxMetadata(zip, metadata, decompressed) {
+  try {
+    const metaPaths = ["meta.xml", "META-INF/meta.xml", "docProps/core.xml"];
+    for (const mp of metaPaths) {
+      const file = zip.file(mp) || Object.values(zip.files).find((f) => f.name.toLowerCase() === mp.toLowerCase()) || null;
+      if (!file) continue;
+      const xml = await file.async("text");
+      if (decompressed) {
+        decompressed.total += xml.length * 2;
+        if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
+      }
+      parseDublinCoreMetadata(xml, metadata);
+      if (metadata.title || metadata.author) return;
+    }
+  } catch {
+  }
+}
+function parseDublinCoreMetadata(xml, metadata) {
+  const parser = createXmlParser();
+  const doc = parser.parseFromString(stripDtd(xml), "text/xml");
+  if (!doc.documentElement) return;
+  const getText = (tagNames) => {
+    for (const tag of tagNames) {
+      const els = doc.getElementsByTagName(tag);
+      if (els.length > 0) {
+        const text = els[0].textContent?.trim();
+        if (text) return text;
+      }
+    }
+    return void 0;
+  };
+  metadata.title = metadata.title || getText(["dc:title", "title"]);
+  metadata.author = metadata.author || getText(["dc:creator", "creator", "cp:lastModifiedBy"]);
+  metadata.description = metadata.description || getText(["dc:description", "description", "dc:subject", "subject"]);
+  metadata.createdAt = metadata.createdAt || getText(["dcterms:created", "meta:creation-date"]);
+  metadata.modifiedAt = metadata.modifiedAt || getText(["dcterms:modified", "meta:date"]);
+  const keywords = getText(["dc:keyword", "cp:keywords", "meta:keyword"]);
+  if (keywords && !metadata.keywords) {
+    metadata.keywords = keywords.split(/[,;]/).map((k) => k.trim()).filter(Boolean);
+  }
+}
+
+// src/hwpx/parser.ts
+async function parseHwpxDocument(buffer, options) {
+  precheckZipSize(buffer, MAX_DECOMPRESS_SIZE, MAX_ZIP_ENTRIES);
+  let zip;
+  try {
+    zip = await JSZip3.loadAsync(buffer);
+  } catch {
+    return extractFromBrokenZip(buffer);
+  }
+  const actualEntryCount = Object.keys(zip.files).length;
+  if (actualEntryCount > MAX_ZIP_ENTRIES) {
+    throw new KordocError("ZIP \uC5D4\uD2B8\uB9AC \uC218 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
+  }
+  const manifestFile = zip.file("META-INF/manifest.xml");
+  if (manifestFile) {
+    const manifestXml = await manifestFile.async("text");
+    if (isEncryptedHwpx(manifestXml)) {
+      if (isComFallbackAvailable() && options?.filePath) {
+        const { pages, pageCount, warnings: warnings2 } = extractTextViaCom(options.filePath);
+        if (pages.some((p) => p && p.trim().length > 0)) {
+          return comResultToParseResult(pages, pageCount, warnings2);
+        }
+      }
+      throw new KordocError("DRM \uC554\uD638\uD654\uB41C HWPX \uD30C\uC77C\uC785\uB2C8\uB2E4. Windows + \uD55C\uCEF4 \uC624\uD53C\uC2A4 \uC124\uCE58 \uC2DC \uC790\uB3D9 \uCD94\uCD9C\uB429\uB2C8\uB2E4.");
+    }
+  }
+  const decompressed = { total: 0 };
+  const metadata = {};
+  await extractHwpxMetadata(zip, metadata, decompressed);
+  const styleMap = await extractHwpxStyles(zip, decompressed);
+  const warnings = [];
+  const sectionPaths = await resolveSectionPaths(zip);
+  if (sectionPaths.length === 0) throw new KordocError("HWPX\uC5D0\uC11C \uC139\uC158 \uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
+  metadata.pageCount = sectionPaths.length;
+  const pageFilter = options?.pages ? parsePageRange(options.pages, sectionPaths.length) : null;
+  const totalTarget = pageFilter ? pageFilter.size : sectionPaths.length;
+  const blocks = [];
+  const shared = createSectionShared();
+  let parsedSections = 0;
+  for (let si = 0; si < sectionPaths.length; si++) {
+    if (pageFilter && !pageFilter.has(si + 1)) continue;
+    const file = zip.file(sectionPaths[si]);
+    if (!file) continue;
+    try {
+      const xml = await file.async("text");
+      decompressed.total += xml.length * 2;
+      if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
+      blocks.push(...parseSectionXml(xml, styleMap, warnings, si + 1, shared));
+      parsedSections++;
+      options?.onProgress?.(parsedSections, totalTarget);
+    } catch (secErr) {
+      if (secErr instanceof KordocError) throw secErr;
+      warnings.push({ page: si + 1, message: `\uC139\uC158 ${si + 1} \uD30C\uC2F1 \uC2E4\uD328: ${secErr instanceof Error ? secErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
+    }
+  }
+  applyPageText(blocks, shared);
+  const images = await extractImagesFromZip(zip, blocks, decompressed, warnings);
+  detectHwpxHeadings(blocks, styleMap);
+  const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
+  const markdown = blocksToMarkdown(blocks);
+  return { markdown, blocks, metadata, outline: outline.length > 0 ? outline : void 0, warnings: warnings.length > 0 ? warnings : void 0, images: images.length > 0 ? images : void 0 };
 }
 
 // src/hwp5/record.ts
@@ -16748,7 +16767,7 @@ function isDistributionSentinel(markdown) {
 }
 
 // src/xlsx/parser.ts
-import JSZip3 from "jszip";
+import JSZip4 from "jszip";
 import { DOMParser as DOMParser2 } from "@xmldom/xmldom";
 var MAX_SHEETS = 100;
 var MAX_DECOMPRESS_SIZE3 = 100 * 1024 * 1024;
@@ -16939,7 +16958,7 @@ function sheetToBlocks(sheetName, grid, merges, maxRow, maxCol, sheetIndex) {
 }
 async function parseXlsxDocument(buffer, options) {
   precheckZipSize(buffer, MAX_DECOMPRESS_SIZE3);
-  const zip = await JSZip3.loadAsync(buffer);
+  const zip = await JSZip4.loadAsync(buffer);
   const warnings = [];
   const workbookFile = zip.file("xl/workbook.xml");
   if (!workbookFile) {
@@ -17645,7 +17664,7 @@ async function parseXlsDocument(buffer, options) {
 }
 
 // src/docx/parser.ts
-import JSZip4 from "jszip";
+import JSZip5 from "jszip";
 import { DOMParser as DOMParser3 } from "@xmldom/xmldom";
 
 // src/docx/equation.ts
@@ -18399,7 +18418,7 @@ async function extractImages(zip, rels, doc, warnings) {
 }
 async function parseDocxDocument(buffer, options) {
   precheckZipSize(buffer, MAX_DECOMPRESS_SIZE4);
-  const zip = await JSZip4.loadAsync(buffer);
+  const zip = await JSZip5.loadAsync(buffer);
   const warnings = [];
   const docFile = zip.file("word/document.xml");
   if (!docFile) {
@@ -19317,7 +19336,7 @@ function fillInlineFields(text, values, filled, matchedLabels) {
 }
 
 // src/form/filler-hwpx.ts
-import JSZip5 from "jszip";
+import JSZip6 from "jszip";
 
 // src/roundtrip/source-map.ts
 function escapeXmlText(text) {
@@ -19991,7 +20010,7 @@ function patchZipEntries(original, replacements) {
 // src/form/filler-hwpx.ts
 async function fillHwpx(hwpxBuffer, values) {
   const u8 = new Uint8Array(hwpxBuffer);
-  const zip = await JSZip5.loadAsync(hwpxBuffer);
+  const zip = await JSZip6.loadAsync(hwpxBuffer);
   const sectionPaths = Object.keys(zip.files).filter((name) => /[Ss]ection\d+\.xml$/i.test(name)).sort();
   if (sectionPaths.length === 0) {
     throw new KordocError("HWPX\uC5D0\uC11C \uC139\uC158 \uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
@@ -20232,7 +20251,7 @@ async function fillHwpx(hwpxBuffer, values) {
 }
 
 // src/hwpx/generator.ts
-import JSZip6 from "jszip";
+import JSZip7 from "jszip";
 
 // src/hwpx/text-metrics.ts
 var ASCII_W = [
@@ -21068,7 +21087,7 @@ async function markdownToHwpx(markdown, options) {
   const gongmunList = gongmun ? precomputeGongmunList(blocks, gongmun) : null;
   const fit = gongmun && gongmunList ? computeGongmunFitPlan(blocks, gongmun, gongmunList) : null;
   const sectionXml = blocksToSectionXml(blocks, theme, gongmun, gongmunList, fit);
-  const zip = new JSZip6();
+  const zip = new JSZip7();
   zip.file("mimetype", "application/hwp+zip", { compression: "STORE" });
   zip.file("META-INF/container.xml", generateContainerXml());
   zip.file("Contents/content.hpf", generateManifest());
@@ -21895,7 +21914,7 @@ function diffTableCells(a, b) {
 }
 
 // src/roundtrip/patcher.ts
-import JSZip7 from "jszip";
+import JSZip8 from "jszip";
 
 // src/roundtrip/table-rows.ts
 var ROW_OBJECT_RE = /<(?:[A-Za-z0-9_]+:)?(?:tbl|pic|equation|ole|container|shape|drawingObject|drawText|video|chart|fieldBegin|fieldEnd|ctrl)\b/;
@@ -22533,7 +22552,7 @@ async function patchHwpx(original, editedMarkdown, options) {
   }
   let zip;
   try {
-    zip = await JSZip7.loadAsync(original);
+    zip = await JSZip8.loadAsync(original);
   } catch {
     return { success: false, applied: 0, skipped, error: "ZIP \uB85C\uB4DC \uC2E4\uD328" };
   }
@@ -23900,10 +23919,10 @@ function stageParaPatch(scan, para, newPlain, skip) {
 }
 
 // src/roundtrip/session.ts
-import JSZip8 from "jszip";
+import JSZip9 from "jszip";
 async function buildState(bytes) {
   const parsed = await parseHwpxDocument(u8ToArrayBuffer2(bytes));
-  const zip = await JSZip8.loadAsync(bytes);
+  const zip = await JSZip9.loadAsync(bytes);
   const sectionPaths = await resolveSectionEntryNames(zip);
   if (sectionPaths.length === 0) {
     throw new Error("HWPX \uC139\uC158 \uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
@@ -24484,7 +24503,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-DCK42RMA.js");
+    const mod = await import("./parser-XEDROIM7.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {